SELF-FERTILIZATION OF PLANTS. 325 
garded as certainly descended from insect-fertilized plants, have regained and retained 
the power of self-fertilization. Thus while Trifolium pratense seems generally to require 
the aid of humble-bees, some very small-flowered species, as 7’. minus, are self-fertile. 
Again, while conspicuous species of Dicentra and Corydalis are adapted for insect 
agency, Fumaria officinalis is highly self-fertile. 
A Both physiologically and morphologically self-sterile plants may become highly self- 
fertile under certain conditions. 
I have enumerated nine proximate causes which appear to be more or less capable of 
accounting for the self-fertilization of certain flowers, or they are at least phenomena cor- 
related with it. Some of these require additional observations to establish them, as a 
sufficient number of cases has not yet been observed; but at all events they are all based 
upon direct observations of usually many more than one instance. 
4. i. The first cause stated is the withering of the corolla. 
Mr. Darwin says of Viola tricolor that a large covered plant set only 18 capsules, 
several of which contained from only one to three seeds; but an uncovered plant set 105 
. fine capsules. 
" The few flowers which produce capsules when insects are excluded are perhaps fertilised by the curl- 
ing inwards of the petals as they wither, for by this means pollen-grains adhering to the papille might 
be inserted into the cavity of the stigma” (J. c. p. 124). 
I can corroborate this fact, for I have found a flower of this plant with the calcarate 
petal ‘glued’ to the stigma with plenty of pollen-grains penetrating the orifice; the 
stamens were detached, and remained suspended above the ovary, just as in cleistogamous 
violets (Tab. XLIV. fig. 6 d). 
In September, 1876, I found a plant of Tradescantia erecta at Kew producing an 
abundance of capsules, but in every instance from unopened flower-buds. The corollas 
were completely shrivelled, and had never expanded, and were matted down upon the 
anthers together with the long style, so that the stigma was impregnated by the pollen 
beneath the withered corolla. Several of the anthers had withered immaturely. I 
gathered several stalks, which, when placed in water, lived for several days, and opened 
their capsules successively. The embryos were perfectly well formed (Tab. XLIV. 
fig. 36). The seeds germinated in 1877, and grew into vigorous plants. Mr. Darwin 
records that Zpomea purpurea is highly self-fertile, while Convolvulus arvensis can 
fertilize itself. `C. sepium closes at night, but not, like C. arvensis, in wet weather. It 
has no smell, and is comparatively little visited by insects. Now all these plants wither 
rapidly, the corolla becoming twisted upon the essential organs; and I strongly suspect 
such a process is highly favourable to the securing of self-fertilization, although the 
stamens do not appear to be proterandrous. The stigmas are somewhat elevated, at 
least at first, beyond the anthers. 
H. Müller also alludes to Hypericum perforatum as being specially assisted in the 
process of self-fertilization by the withering of the corolla. He observes that the 
anthers and stigmas mature together, and that although there is no honey, yet the 
conspicuousness of the flowers will attract pollen-seeking insects which might cross 
SECOND SERIES.—BOTANY, VOL. I. 22 
